README.md

---
title: Sentiment Analysis
emoji: 🎙
colorFrom: purple
colorTo: green
sdk: gradio
sdk_version: "5.0.0"
python_version: "3.10"
app_file: app.py
pinned: false
---

# AI Conversation Sentiment Analyzer

Upload or record audio to get speaker-separated transcription with sentiment labels.

**Models used:**
- openai/whisper-base — transcription
- pyannote/speaker-diarization-3.1 — speaker diarization
- nlptown/bert-base-multilingual-uncased-sentiment — sentiment analysis

**Required secret:** Set `HF_TOKEN` under Space Settings → Repository secrets.